Meta, the Mark Zuckerberg company that owns and operates Facebook, Instagram, and WhatsApp, and other, has announced plans to release an artificial intelligence system to rival other AI models, the Wall Street Journal reported on Sunday.
Meta has already released an AI model, Llama 2, which came out earlier this year. The next one, the company says, will be multiple times more powerful.
Initial plans for the Meta AI platform aim to have the system be of utility to other companies who may use it for tasks such as generating text, creating services, analysis, and other such things on a highly sophisticated level.
With this in mind, Zuckerberg is looking to make the Meta AI open-sourced.
Zuckerberg spearheaded the project, bringing together the group that is developing the AI system, known as a large language model.
The Wall Street Journal further reported that some of those familiar with the project said that Meta would begin training the model in early 2024.
The Meta AI system is the technology giant’s attempt to become a serious competitor in the AI world after companies such as Open AI have dominated the market over the past months.
Training the Meta AI
The Wall Street Journal reported that Mata is gearing up for a significant effort at producing its own open-sourced AI platform, building up data centers and obtaining the best Nvidia chips, H100s, to use in training the AI.
Additionally, as opposed to its development of Llama 2, which was done in conjunction with Microsoft, the new AI project is said to be done independently by Meta.
The new Meta AI aims to be roughly as capable as GPT-4, which OpenAI launched in March.
The Wall Street Journal noted that Zuckerberg will be one of the tech company executives to attend a summit organized by Senate Majority Leader Chuck Schumer this coming Wednesday to discuss the future of AI.